Splitting statistics

This code analyses splitting statistics for CTC-clusters.

The analysis takes a list of trees sampled from its posterior distribution as input and samples mutations placements for each of the trees.

Configure the script

inputFolder <- "/Users/jgawron/Documents/projects/CTC_backup/input_folder"
simulationInputFolder <- "/Users/jgawron/Documents/projects/CTC_backup/simulations/simulations2"
treeName <- "LM2"
nTreeSamplingEvents <- 1000
nMutationSamplingEvents <- 1000

Loading data

source("/Users/jgawron/Documents/projects/CTC-SCITE/CTC-SCITE/experiments/workflow/resources/functions.R")
## ── Attaching core tidyverse packages ─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr     1.1.4     ✔ readr     2.1.5
## ✔ forcats   1.0.0     ✔ stringr   1.5.1
## ✔ ggplot2   3.5.1     ✔ tibble    3.2.1
## ✔ lubridate 1.9.3     ✔ tidyr     1.3.1
## ✔ purrr     1.0.2     
## ── Conflicts ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
input <- load_data(inputFolder, treeName)
## Rows: 40000 Columns: 5
## ── Column specification ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (1): Tree
## dbl (4): LogScore, SequencingErrorRate, DropoutRate, LogTau
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 883 Columns: 72
## ── Column specification ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr  (3): X1, X3, X4
## dbl (69): X2, X5, X6, X7, X8, X9, X10, X11, X12, X13, X14, X15, X16, X17, X1...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 34 Columns: 5
## ── Column specification ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (2): Cluster, Description
## dbl (3): CellCount, TCs, WBCs
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
postSampling <- input$postSampling
nClusters <- input$nClusters
ClusterID <- input$clusterID
nCells <- input$nCells  
nMutations <- input$nMutations
nClusters <- input$nClusters
alleleCount <- input$alleleCount
mutatedReadCounts <- input$mutatedReadCounts
totalReadCounts <- input$totalReadCounts
sampleDescription <- input$sample_description

Sample description

Each row corresponds to a cell. Column description: - Cluster: An number indicating which sample the cell belongs to. - ClusterName: The name of the sample in the nodeDescription.tsv file - WBC: a binary vector indicating whether the cell is a white blood cell (1) or not (0). - color: Indicates the color of the cluster in the tree, as described in the nodeDescription.tsv file.

print(sampleDescription)
##    Cluster ClusterName WBC          color single_cell
## 1        0     LM2_A81   0         gray93        TRUE
## 2        1     LM2_A83   0         gray93        TRUE
## 3        2     LM2_A85   0         gray93        TRUE
## 4        3     LM2_A86   0     lightcoral       FALSE
## 5        3     LM2_A86   0     lightcoral       FALSE
## 6        4     LM2_A89   0         gray93        TRUE
## 7        5     LM2_A90   0         gray93        TRUE
## 8        6     LM2_A93   0     sandybrown       FALSE
## 9        6     LM2_A93   0     sandybrown       FALSE
## 10       7     LM2_A94   0       skyblue3       FALSE
## 11       7     LM2_A94   0       skyblue3       FALSE
## 12       8 LM2_CTC_G19   0        thistle        TRUE
## 13       9 LM2_CTC_G21   0        thistle        TRUE
## 14      10 LM2_CTC_G32   0   lemonchiffon        TRUE
## 15      11 LM2_CTC_G33   0   lemonchiffon        TRUE
## 16      12 LM2_CTC_G34   0   lemonchiffon        TRUE
## 17      13      LM2_E1   0         gray93        TRUE
## 18      14      LM2_E2   0         gray93        TRUE
## 19      15      LM2_E3   0         gray93        TRUE
## 20      16      LM2_E5   0     violetred3       FALSE
## 21      16      LM2_E5   0     violetred3       FALSE
## 22      17      LM2_E6   0         gray93        TRUE
## 23      18      LM2_E7   0 lightslateblue       FALSE
## 24      18      LM2_E7   0 lightslateblue       FALSE
## 25      19      LM2_G1   0 paleturquoise3       FALSE
## 26      19      LM2_G1   0 paleturquoise3       FALSE
## 27      20     LM2_G10   0         khaki3       FALSE
## 28      20     LM2_G10   0         khaki3       FALSE
## 29      21     LM2_G11   0         gray93        TRUE
## 30      22     LM2_G12   0         gray93        TRUE
## 31      23     LM2_G13   0         gray93        TRUE
## 32      24     LM2_G14   0  darkseagreen4       FALSE
## 33      24     LM2_G14   0  darkseagreen4       FALSE
## 34      25     LM2_G15   0           gold       FALSE
## 35      25     LM2_G15   0           gold       FALSE
## 36      25     LM2_G15   0           gold       FALSE
## 37      25     LM2_G15   0           gold       FALSE
## 38      25     LM2_G15   0           gold       FALSE
## 39      26      LM2_G2   0           plum        TRUE
## 40      27      LM2_G3   1           plum       FALSE
## 41      27      LM2_G3   0           plum       FALSE
## 42      27      LM2_G3   0           plum       FALSE
## 43      28      LM2_G4   0    yellowgreen        TRUE
## 44      29      LM2_G5   0    yellowgreen        TRUE
## 45      30      LM2_G6   0    yellowgreen        TRUE
## 46      31      LM2_G7   0    yellowgreen        TRUE
## 47      32      LM2_G8   1    yellowgreen       FALSE
## 48      32      LM2_G8   0    yellowgreen       FALSE
## 49      33      LM2_G9   0   navajowhite2       FALSE
## 50      33      LM2_G9   0   navajowhite2       FALSE

Get null distributions of relevant statistics, stratified by sample:

cutoffsSplittingProbs <- data.frame(clusterSize = vector(), Cutoff = vector())
cutoffsBranchingProbabilities <- data.frame(clusterSize = vector(), Cutoff = vector())

for (clusterSize in 2:5){
  try(
  {treeNameSimulated <- paste(treeName, clusterSize, sep = '_')


  inputSimulated <- load_data(simulationInputFolder, treeNameSimulated)

  postSamplingSimulated <- inputSimulated$postSampling
  nClustersSimulated <- inputSimulated$nClusters
  ClusterIDSimulated <- inputSimulated$clusterID
  nCellsSimulated <- inputSimulated$nCells  
  nMutationsSimulated <- inputSimulated$nMutations
  nClustersSimulated <- inputSimulated$nClusters
  alleleCountSimulated <- inputSimulated$alleleCount
  mutatedReadCountsSimulated <- inputSimulated$mutatedReadCounts
  totalReadCountsSimulated <- inputSimulated$totalReadCounts
  sampleDescriptionSimulated <- inputSimulated$sample_description
  
  distance <- computeClusterSplits(sampleDescriptionSimulated, postSamplingSimulated, treeNameSimulated, nCellsSimulated,
                     nMutationsSimulated, nClustersSimulated,
                     alleleCountSimulated,
                     mutatedReadCountsSimulated, totalReadCountsSimulated,
                     nMutationSamplingEvents = nMutationSamplingEvents, nTreeSamplingEvents = nTreeSamplingEvents,
                     cellPairSelection = c("orchid", "orchid1", "orchid2",
                                           "orchid3", "orchid4", "darkorchid",
                                           "darkorchid1","darkorchid2", "darkorchid3",
                                           "darkorchid4", "purple", "purple1",
                                           "purple2", "purple3", "purple4"))

  

  plot(ggplot(distance$splittingProbs, aes(x = "Values", y = Splitting_probability, fill = 'Splitting_probability')) +
    geom_boxplot())
  cutoffsSplittingProbs <- rbind(cutoffsSplittingProbs, data.frame(clusterSize = clusterSize, Cutoff = mean(distance$splittingProbs$Splitting_probability) + 2 * sd(distance$splittingProbs$Splitting_probability) ))
  
  ##Note that the way the aggregatedBranchingProbabilities are computed all pairs of cells from the same cluster are
  ## taken into account. This has the effect that clusters with more cells would be counted more often and contribute more
  ## to the shape of the final distribution. This is no problem right now as we only aggregate counts from clusters
  ## of the same size, it is however the potential source of a future bug!!
  
  plot(ggplot(data.frame(x = distance$aggregatedBranchingProbabilities), aes(x = x)) +
    geom_histogram(binwidth = 0.01))
  print(data.frame(clusterSize = clusterSize, Cutoff = quantile(distance$aggregatedBranchingProbabilities, probs = 0.95, names = FALSE)[1] ))
  cutoffsBranchingProbabilities <- rbind(cutoffsBranchingProbabilities, data.frame(clusterSize = clusterSize, Cutoff = quantile(distance$aggregatedBranchingProbabilities, probs = 0.95, names = FALSE)[1] ))
  })
}
## Rows: 65960 Columns: 5
## ── Column specification ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (1): Tree
## dbl (4): LogScore, SequencingErrorRate, DropoutRate, LogTau
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 883 Columns: 80
## ── Column specification ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr  (3): X1, X3, X4
## dbl (77): X2, X5, X6, X7, X8, X9, X10, X11, X12, X13, X14, X15, X16, X17, X1...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 38 Columns: 5
## ── Column specification ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (2): Cluster, Description
## dbl (3): CellCount, TCs, WBCs
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## [1] "Computing genomic distances of leaves: 51 50"
## [1] "Computing the posterior distribution"

## [1] "Computing genomic distances of leaves: 53 52"
## [1] "Computing the posterior distribution"

## [1] "Computing genomic distances of leaves: 55 54"
## [1] "Computing the posterior distribution"

## [1] "Computing genomic distances of leaves: 57 56"
## [1] "Computing the posterior distribution"

##   clusterSize       Cutoff
## 1           2 6.306453e-06
## Rows: 25118 Columns: 5
## ── Column specification ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (1): Tree
## dbl (4): LogScore, SequencingErrorRate, DropoutRate, LogTau
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 883 Columns: 78
## ── Column specification ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr  (3): X1, X3, X4
## dbl (75): X2, X5, X6, X7, X8, X9, X10, X11, X12, X13, X14, X15, X16, X17, X1...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 37 Columns: 5
## ── Column specification ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (2): Cluster, Description
## dbl (3): CellCount, TCs, WBCs
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

## [1] "Computing genomic distances of leaves: 51 50"
## [1] "Computing the posterior distribution"

## [1] "Computing genomic distances of leaves: 54 53"
## [1] "Computing the posterior distribution"

## [1] "Computing genomic distances of leaves: 57 56"
## [1] "Computing the posterior distribution"

##   clusterSize      Cutoff
## 1           3 3.75624e-06
## Rows: 24534 Columns: 5
## ── Column specification ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (1): Tree
## dbl (4): LogScore, SequencingErrorRate, DropoutRate, LogTau
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 883 Columns: 76
## ── Column specification ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr  (3): X1, X3, X4
## dbl (73): X2, X5, X6, X7, X8, X9, X10, X11, X12, X13, X14, X15, X16, X17, X1...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 36 Columns: 5
## ── Column specification ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (2): Cluster, Description
## dbl (3): CellCount, TCs, WBCs
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

## [1] "Computing genomic distances of leaves: 51 50"
## [1] "Computing the posterior distribution"

## [1] "Computing genomic distances of leaves: 55 54"
## [1] "Computing the posterior distribution"

##   clusterSize       Cutoff
## 1           4 2.807941e-06
## Warning: One or more parsing issues, call `problems()` on your data frame for details, e.g.:
##   dat <- vroom(...)
##   problems(dat)
## Rows: 54437 Columns: 5
## ── Column specification ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (1): Tree
## dbl (4): LogScore, SequencingErrorRate, DropoutRate, LogTau
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 883 Columns: 76
## ── Column specification ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr  (3): X1, X3, X4
## dbl (73): X2, X5, X6, X7, X8, X9, X10, X11, X12, X13, X14, X15, X16, X17, X1...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 36 Columns: 5
## ── Column specification ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## Delimiter: "\t"
## chr (2): Cluster, Description
## dbl (3): CellCount, TCs, WBCs
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

## [1] "Computing genomic distances of leaves: 51 50"
## [1] "Computing the posterior distribution"

## [1] "Computing genomic distances of leaves: 56 55"
## [1] "Computing the posterior distribution"

##   clusterSize       Cutoff
## 1           5 3.219289e-06

Get the relevant statistics for each of the clusters of a dataset and output numbers of oligoclonal clusters:

nTumorClusters <- 0
nOligoclonalClusters1 <- 0
nOligoclonalClusters2 <- 0
splittingSummary1 <- data.frame(Color = vector(), Oligoclonal = vector(), ClusterSize = vector())
splittingSummary2 <- data.frame(Color = vector(), Oligoclonal = vector(), ClusterSize = vector())

for(clusterSize in 2:5){
  try({
    clusterColor <- sampleDescription %>%
    filter(WBC ==0 &  color != 'gray93') %>%
    group_by(color) %>%
    filter(n() == clusterSize) %>%
    pull(color) %>%
    unique() 
    
    for(color in clusterColor){
      distance <- computeClusterSplits(sampleDescription, postSampling, treeName, nCells,
                     nMutations, nClusters,
                     alleleCount,
                     mutatedReadCounts, totalReadCounts,
                     nMutationSamplingEvents = nMutationSamplingEvents, nTreeSamplingEvents = nTreeSamplingEvents,
                     cellPairSelection = c(color))

      splittingProbs <- mean(distance$splittingProbs$Splitting_probability)
      branchingProbs <- mean(distance$aggregatedBranchingProbabilities)
    
      nTumorClusters <- nTumorClusters + 1
      oligoclonal <- FALSE
      print(clusterSize)
      print(cutoffsSplittingProbs[(cutoffsSplittingProbs$clusterSize == clusterSize), 2])
      if(splittingProbs > (cutoffsSplittingProbs[(cutoffsSplittingProbs$clusterSize == clusterSize), 2])){
        nOligoclonalClusters1 <- nOligoclonalClusters1 + 1
        oligoclonal <- TRUE
      }
      splittingSummary1 <- rbind(splittingSummary1, data.frame(Color = color, Oligoclonal = oligoclonal, ClusterSize = clusterSize))
      oligoclonal <- FALSE
      if(branchingProbs > cutoffsBranchingProbabilities[(cutoffsBranchingProbabilities$clusterSize == clusterSize), 2]){
        nOligoclonalClusters2 <- nOligoclonalClusters2 + 1
        oligoclonal <- TRUE
      }
      splittingSummary2 <- rbind(splittingSummary2, data.frame(Color = color, Oligoclonal = oligoclonal, ClusterSize = clusterSize))
    }
  })
}
## [1] "Computing genomic distances of leaves: 4 3"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 8 7"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 10 9"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 12 11"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 20 19"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 23 22"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 25 24"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 27 26"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 32 31"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 49 48"
## [1] "Computing the posterior distribution"

## [1] 2
## [1] 0.08619833
## [1] "Computing genomic distances of leaves: 14 13"
## [1] "Computing the posterior distribution"

## [1] 3
## [1] 0.007393762
## [1] "Computing genomic distances of leaves: 40 38"
## [1] "Computing the posterior distribution"

## [1] 3
## [1] 0.007393762
## [1] "Computing genomic distances of leaves: 34 33"
## [1] "Computing the posterior distribution"

## [1] 5
## [1] 0.01719647
## [1] "Computing genomic distances of leaves: 43 42"
## [1] "Computing the posterior distribution"

## [1] 5
## [1] 0.01719647
numberOfCancerClusters <- sampleDescription %>%
    filter(WBC ==0 &  color != 'gray93') %>%
    group_by(color) %>%
    filter(n() > 1) %>%
    pull(color) %>%
    unique() %>% length() 

print(sprintf('%d out of %d clusters were found to be oligoclonal in %s, using method 1', nOligoclonalClusters1, numberOfCancerClusters, treeName))
## [1] "8 out of 14 clusters were found to be oligoclonal in LM2, using method 1"
print(sprintf('%d out of %d clusters were found to be oligoclonal in %s, using method 2', nOligoclonalClusters2, numberOfCancerClusters, treeName))
## [1] "11 out of 14 clusters were found to be oligoclonal in LM2, using method 2"
print(splittingSummary1)
##             Color Oligoclonal ClusterSize
## 1      lightcoral       FALSE           2
## 2      sandybrown       FALSE           2
## 3        skyblue3       FALSE           2
## 4         thistle        TRUE           2
## 5      violetred3        TRUE           2
## 6  lightslateblue       FALSE           2
## 7  paleturquoise3        TRUE           2
## 8          khaki3       FALSE           2
## 9   darkseagreen4        TRUE           2
## 10   navajowhite2       FALSE           2
## 11   lemonchiffon        TRUE           3
## 12           plum        TRUE           3
## 13           gold        TRUE           5
## 14    yellowgreen        TRUE           5
print(splittingSummary2)
##             Color Oligoclonal ClusterSize
## 1      lightcoral       FALSE           2
## 2      sandybrown       FALSE           2
## 3        skyblue3       FALSE           2
## 4         thistle        TRUE           2
## 5      violetred3        TRUE           2
## 6  lightslateblue        TRUE           2
## 7  paleturquoise3        TRUE           2
## 8          khaki3        TRUE           2
## 9   darkseagreen4        TRUE           2
## 10   navajowhite2        TRUE           2
## 11   lemonchiffon        TRUE           3
## 12           plum        TRUE           3
## 13           gold        TRUE           5
## 14    yellowgreen        TRUE           5